Challenges in managing semantic annotations in harvested research objects in a national CRIS context

Harvested metadata on research objects can include links between the primary domain such as organizational identifiers associated with dataset, persons identified ORCIDs linked to publications and connected through ISSNs publishing channels. This kind of linkage is bread-and-butter CRIS systems usually comprehensively maintained. When it comes more subjective description a object, keywords, themes, or subject headings, issues related data management modeling become prominent challenges flexibility free text keywords opposed authoritative, but rigid classification systems. Many also already contain an extensive content, just meant for human consumption, in form abstract similar summary text.With help automated mining annotation tools, these textual representations be processed into structured data. paper presents processing pipelines implemented part research.fi portal automatic linking different inputs based automatically extracted ontology concepts discusses implications utilizing them platform. But than simply discussing creation word clusters representation semantic content objects, we discuss maintaining produced